|Publication number||US7627475 B2|
|Application number||US 11/716,240|
|Publication date||Dec 1, 2009|
|Priority date||Aug 31, 1999|
|Also published as||US7222075, US20020194002, US20070162283|
|Publication number||11716240, 716240, US 7627475 B2, US 7627475B2, US-B2-7627475, US7627475 B2, US7627475B2|
|Inventors||Valery A. Petrushin|
|Original Assignee||Accenture Llp|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (61), Non-Patent Citations (37), Referenced by (35), Classifications (9), Legal Events (4)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This is a continuation of U.S. patent application Ser. No. 10/194,908, filed Jul. 12, 2002, now U.S. Pat. No. 7,222,075, which is a continuation-in-part of U.S. application Ser. No. 09/833,301, filed Apr. 10, 2001, which is a continuation of U.S. patent application Ser. No. 09/388,909, filed Aug. 31, 1999, now U.S. Pat. No. 6,275,806, which are herein incorporated by reference in their entirety.
The present invention relates to analysis of speech and more particularly to detecting emotion using statistics and neural networks to classify speech signal parameters according to emotions the networks have been taught to recognize.
Although the first monograph on expression of emotions in animals and humans was written by Charles Darwin in the nineteenth century and psychologists have gradually accumulated knowledge in the field of emotion detection and voice recognition, it has attracted a new wave of interest recently by both psychologists and artificial intelligence specialists. There are several reasons for this renewed interest, including technological progress in recording, storing and processing audio and visual information, the development of non-intrusive sensors, the advent of wearable computers; and the urge to enrich human-computer interfaces from point-and-click to sense-and-feel. Further, a new field of research in Artificial Intelligence (AI) known as affective computing has recently been identified. Affective computing focuses research on computers and emotional states, combining information about human emotions with computing power to improve human-computer relationships.
As to research on recognizing emotions in speech, psychologists have done many experiments and suggested many theories. In addition, AI researchers have made contributions in the areas of emotional speech synthesis, recognition of emotions, and the use of agents for decoding and expressing emotions.
A closer look at how well people can recognize and portray emotions in speech is revealed in Tables 1-4. Thirty subjects of both genders recorded four short sentences with five different emotions (happiness, anger, sadness, fear, and neutral state or normal). Table 1 shows a performance confusion matrix, in which only the numbers on the diagonal match the intended (true) emotion with the detected (evaluated) emotion. The rows and the columns represent true and evaluated categories respectively. For example, the second row indicates that 11.9% of utterances that were portrayed as happy were evaluated as neutral (unemotional), 61.4% as truly happy, 10.1% as angry, 4.1% as sad, and 12.5% as afraid. The most easily recognizable category is anger (72.2%) and the least recognizable category is fear (49.5%). There is considerable confusion between sadness and fear, sadness and unemotional state, and happiness and fear. The mean accuracy of 63.5% (diagonal numbers divided by five) agrees with results of other experimental studies.
Performance Confusion Matrix
Table 2 shows statistics for evaluators for each emotional category and for summarized performance that was calculated as the sum of performances for each category. It can be seen that the variance for anger and sadness is much less then for the other emotional categories.
Table 3 below shows statistics for “actors”, i.e. how well subjects portray emotions. Speaking more precisely, the table shows how readily a particular portrayed emotion is recognized by evaluators. It is interesting to compare tables 2 and 3 and see that the ability to portray emotions (total mean is 62.9%) at about the same level as the ability to recognize emotions (total mean is 63.2%). However, the variance for portraying and emotion is much larger.
Table 4 shows self-reference statistics, i.e. how well subjects were able to recognize their own portrayals. We can see that people do much better in recognizing their own emotions (mean is 80.0%), especially for anger (98.1%), sadness (80.0%) and fear (78.8%). Interestingly, fear was recognized better than happiness. Some subjects failed to recognize their own portrayals for happiness and the normal or neutral state.
These results provide valuable insight about human performance and can serve as a baseline for comparison to computer performance. In spite of the research on recognizing emotions in speech, little has been done to provide methods and apparatuses that utilize emotion recognition for business purposes.
One embodiment of the present invention is a method of detecting emotional states. The method comprises providing speech signals, dividing the speech signals into at least one of segments, frames and subframes. The method also includes extracting acoustic features from the speech signals, and calculating statistics from the acoustic features. The statistics serve as inputs to a classifier, which can be represented as a computer program, a device or a combination of both. The method also includes classifying speech signals with at least one neural network classifier as belonging to at least one emotional state, and storing the speech signals and the emotional states in a storage medium, in a manner to allow later retrieval of the stored speech signals and emotional states. The method also includes outputting an indication of the at least one emotional state in a human-recognizable format. The at least one neural network classifier is taught to recognize at least one emotional state from a finite number of emotional states.
Another embodiment of the invention is a system for classifying speech. The system comprises a computer system having a central processing unit (CPU), an input device, at least one memory for storing data indicative of speech signals, and an output device. The computer system also comprises logic for receiving and analyzing speech signals, logic for dividing the speech signals, and logic for extracting at least one feature from the speech signals. The system comprises logic for calculating statistics of the speech, and logic for at least one neural network for classifying the speech as belonging to at least one of a finite number of emotional states. The system also comprises logic for storing the speech signals and the emotional states in a storage medium, in a manner to allow later retrieval of the stored speech signals and emotional states, and logic for outputting an indication of the at least one emotional state.
Another embodiment of the invention is a system for detecting emotional states in a voice signal. The system comprises a speech reception device, and at least one computer connected to the speech reception device. The system further comprises at least one memory operably connected to the at least one computer, and a computer program including at least one neural network for dividing the voice signals into a plurality of segments, and for analyzing the voice signals according to features of the segments to detect the emotional states in the voice signals. The system also comprises a database of speech signal features and statistics accessible to the computer for comparison with features of the voice signals, and an output device coupled to the computer for notifying a user of the emotional state detected in the voice signals.
These and many other aspects of the invention will become apparent through the following drawings and detailed description of embodiments of the invention, which are meant to illustrate, but not the limit the embodiments thereof.
The invention will be better understood when consideration is given to the following, detailed description thereof. Such description makes reference to the attached drawings wherein:
The present invention is directed towards recognizing emotions in speech, which may have useful and valuable applications for business purposes. Recognizing emotions may help call-center personnel deal with angry or emotional callers. Knowing a customer or caller's emotional state may help operators deal with callers who are angry or excited. Conversely, detecting little emotion in a caller in whom excitement or happiness is expected may also prove useful. Detecting other emotions, such as nervousness or fear, may alert businesses to persons who may be attempting to cheat or defraud them. There are many business uses for a system or a method that detects emotions in persons.
Some embodiments of the present invention may be used to detect the emotion of a person based on a voice analysis and to output the detected emotion of the person. Other embodiments of the present invention may be used for the detecting the emotional state of a caller in telephone call center conversations, and for providing feedback to an operator or a supervisor for monitoring purposes. Other embodiments of the present invention may be used for classifying voice mail messages according to the emotions expressed by a caller. Yet other embodiments of the present invention may be used for emotional training of several categories of people, including call center operators, would-be dramatic actors, and people suffering from autism. Another area of application for embodiments of the present invention is in detecting nervousness or fear in a business environment.
In accordance with at least one embodiment of the present invention, a system is provided for voice processing and analysis. The system may be enabled using a hardware implementation such as that illustrated in
A representative hardware environment of a preferred embodiment of the present invention is depicted in
The present invention uses a data-driven approach for creating an emotion recognition system. This choice is motivated by knowing the complexity of the emotional expression among different languages, cultural traditions, and age differences among targeted users. Moreover, the characteristics of a speech signal are heavily dependent on the equipment and procedures used for the acquisition and digitizing of the speech signal.
Steps in creating the emotion recognition system are depicted in
The data collection stage includes the steps depicted in
The stage of creating a classifier consists of the steps shown in
The system development stage includes the following steps. The classifier is embedded into a system using interfaces. In the embodiments of the present invention, the process depicted in
In one aspect of the present invention, the classifier includes probabilities of particular voice features being associated with an emotion. Preferably, the selection of the emotion using the classifier includes analyzing the probabilities and selecting the most probable emotion based on the probabilities. Optionally, the probabilities may include performance confusion statistics, such as are shown in the performance confusion matrix, above in Table 1. Additionally, the statistics may include self-recognition statistics, as shown above in Table 4.
Partitioning the Speech Signal
To train and test a classifier, the input speech signal is partitioned into fragments or segments, which ideally should correspond to phrases. Experimental research has demonstrated that phrases of conversational English have a length of from 1 second to 3 seconds. An algorithm for partitioning the speech signal into segments uses energy values to detect speech segments and select phrases. The algorithm works in the following manner. First, an energy value is calculated for each fragment of a length of 20 milliseconds. Then the values are compared to a threshold to detect speech segments. A median filter is applied to the resulting binary vector to smooth the vector. After this step, the algorithm finds the beginning of a speech signal and considers a speech segment of length 4 seconds starting from this point. For this segment, the largest pause lying in the interval from 1 second to 3 seconds is detected and the segment is cut at this pause. If no pauses are found then, a segment 3 seconds long is selected. The process continues for the rest of the signal. The signals may be further divided into frames, typically from about 20 to about 40 milliseconds long, and subframes, typically about 10 to about 20 milliseconds long. Other lengths of time, longer or shorter, may be used for segments, frames and subframes.
It has been found that pitch is the main vocal cue for emotion recognition. Pitch is represented by the fundamental frequency (F0) of the speech sample, i.e. the lowest frequency of the vibration of the vocal folds. Other acoustic variables contributing to vocal emotion signaling include the following: energy or amplitude of the speech signal; frequency spectrum; formants and temporal features, such as duration; and pausing. Another approach to feature extraction is to enrich the set of features by considering some derivative features, such as the linear predictive coding (LPC) cepstrum coefficients, mel-frequency cepstrum coefficients (MFCC) or features of the smoothed pitch contour and its derivatives. In experimental work, the features of prosodic (suprasegmental) acoustic features, such as fundamental frequency, duration, formants, and energy were used.
There are several approaches to calculating F0. In one of the embodiments of this invention, a variant of the approach proposed by Paul Boersma was used. More details on the algorithm are set forth in the publication Proc. Inst. for Phonetic Sciences, University of Amsterdam, vol. 17 (1993), pp. 97-110, in an article by Paul Boersma entitled, “Accurate Short-Term Analysis of the Fundamental Frequency and the Harmonic-to-Noise Ratio of a Sampled Sound,” which is herein incorporated by reference. To calculate the fundamental frequency the speech signal is divided into a plurality of overlapped frames. Each frame is 40 milliseconds long and the next frame overlaps the previous one by 30 milliseconds. The fundamental frequency is calculated only for the voiced part of an utterance. Additionally, for F0 the slope can be calculated as a linear regression for the voiced part of speech, i.e. the line that fits the pitch contour. Subframes may be selected to have one or more lengths.
Formants are the resonances of the vocal tract. Their frequencies are higher than the basic frequency. The formants are enumerated in ascending order of their frequencies. For one of the embodiments of this invention, the first three formants (F1, F2, and F3) and their bandwidths (BW1, BW2, and BW3) were estimated using an approach based on picking peaks in the smoothed spectrum obtained by LPC analysis, and solving for the roots of a linear predictor polynomial. Formants are calculated for each 20-millisecond subframe, overlapped by 10 milliseconds. Energy is calculated for each 10-millisecond subframe as a square root of the sum of squared samples. The relative voiced energy can also be calculated as the proportion of voiced energy to the total energy of utterance. The speaking rate can be calculated as the inverse of the average length of the voiced part of utterance.
For a number of voice features, the following statistics can be calculated: mean, standard deviation, minimum, maximum and range. Statistic selection algorithms can be used to estimate the importance of each statistic. In experimental work, the RELIEF-F algorithm was used for selection. The RELIEF-F has been run for the data set, varying the number of nearest-neighbors from 1 to 12, and the features ordered according to their sum of ranks. The top 14 statistics are the following: F0 maximum, F0 standard deviation, F0 range, F0 mean, BW1 mean, BW2 mean, energy standard deviation, speaking rate, F0 slope, F1 maximum, energy maximum, energy range, F2 range, and F1 range. To investigate how sets of statistics influence the accuracy of emotion recognition algorithms, three nested sets of statistics may be formed based on their sum of ranks. The first set includes the top eight statistics (from F0 to maximum speaking rate), the second set extends the first set by the two next statistics (F0 slope and F1 maximum), and the third set includes all 14 top statistics. More details on the RELIEF-F algorithm are set forth in the publication Proc. European Conf. On Machine Learning (1994), pp. 171-182, in the article by I. Kononenko entitled, “Estimating attributes: Analysis and extension of RELIEF,” which is herein incorporated by reference.
A number of models may be used to create classifiers for recognizing emotion in speech. In experimental work for the present invention, the following models have been used: nearest neighbor, back propagation neural networks, and ensembles of classifiers. The input vector to a classifier consists of 8, 10 or 14 elements or statistics, depending on the set of elements used. The vector input to a classifier thus may consist of 8 statistics, including a maximum value of a fundamental frequency, a standard deviation of the fundamental frequency, a range of the fundamental frequency, a mean of the fundamental frequency, a mean of a bandwidth of a first formant, a mean of a bandwidth of a second formant, an energy standard deviation, and a speaking rate. If the vector input to a classifier consists of ten elements or statistics, they may include the above eight, and in addition, a slope of a fundamental frequency and a maximum of a first formant. Finally, if the vector input classifier consists of fourteen statistics, they may include the above ten statistics, and in addition, an energy maximum, an energy range, a first formant range, and a second formant range.
A two-layer back propagation neural network architecture was used to create neural network classifiers. A classifier has an 8-, 10- or 14-element (statistics) input vector, with 10 or 20 nodes in the hidden sigmoid layer and five nodes in the output linear layer. The number of outputs corresponds to the number of emotional categories. Several neural network classifiers were trained on the training data set using different initial weight matrices for the neural network. This approach, when applied to the test data set and the 8-statistic set above, gave an average accuracy of about 65% with the following distribution for emotional categories: normal state, 55-65%; happiness, 60-70%; anger, 60-80%; sadness, 60-70%; and fear, 25-50%.
Ensembles of neural network classifiers also have been used. An ensemble consists of an odd number of neural network classifiers, which have been trained on different subsets of the training set using the bootstrap aggregation and cross-validated committee techniques. Bootstrap aggregation involves taking a number of “bootstrap” replicates of the training set and deriving from each one classification predictions for the entire test set and averaging them over all the bootstrap replicates. Another technique that has proven useful is the use of “cross-validated committees.” In this technique, overlapping training sets may be constructed by leaving out a different feature or parameter in each set. The sets so constructed are then compared. The ensemble makes decisions based on a majority voting principle. Suggested ensemble sizes are from 7 to 25.
The last approach is based on the following idea. Instead of training a neural network to recognize all emotions, build a set of specialists or experts that can recognize only one emotion and then combine their results to classify a given sample. To train the experts, a two-layer back-propagation neural network architecture was used. This architecture has an 8-element input vector, 10 or 20 nodes in the hidden sigmoid layer, and one node in the output linear layer. The same training and test sets were used but with only two classes (for example, angry and non-angry).
The important question is how to combine opinions of the experts to classify a given sample. A simple and natural rule is to choose the class in which the expert's value is closest to unity. This rule gives an accuracy of about 60% for the 10-neuron architecture and about 53% for the 20-neuron architecture (
In general, the approach that outperformed the others was based on ensembles of neural network recognizers. This approach was chosen for the embodiments described below.
Exemplary Apparatuses for Detecting Emotion in Voice Signals
This section describes several apparatuses for analyzing speech in accordance with the present invention and their application for business purposes.
Voice Messaging System
In a call center environment, an agent may be assigned to call back for messages with a particular emotional content, for example for messages with dominant negative emotions, e.g., sadness, anger or fear. A speech recognition engine can be applied to the message to obtain a transcript as an additional annotation. The annotations and decisions 1160 can be saved and the results output 1170. The output may take the form of a signal or message on a computer, a printed message from a printer, a video display or output device connected to a computer, an audible signal or tone output from an audio output device, or even an alarm. The output may also be routed to predetermined locations based on the emotional content of the message. Routings may include a voice-mail system, an e-mail system or destination, a call center, a customer service center, a manager, or even emergency response personnel.
There are many different ways in which determined emotions can be presented to users in a human-readable (i.e., human recognizable, audible or visual) format. These examples are intended to illustrate and not intended to limit the invention. In a call center application, the summary of the system operation can be presented as an electronic or paper document that summarizes the emotional content of each message, the telephone number to call back, the transcript of the message, and the name of the person assigned to call back. In an application that is designed for managing personal voice mail messages, the system can include additional information to indicate the emotional content of the messages.
For a telephone-based solution, for example, the system can add the following message, “You have three new messages, two of them are highly emotional. Press 1, if you want to listen to the emotional messages first.” For a computer-based solution, for example, the system can assign a pictogram or icon that represents the emotional content of the message (an “emoticon”) to each message in the mailbox and the system can sort the personal voice mail messages according to their emotional content on request from the user. In the case of a meeting, where a participant or an observer desires to know the emotional state of the other persons present, a signal may be given in a human-recognizable manner, such as by flashing a light or a visible signal, by sounding a tone, or by displaying an icon or message on a computer accessible to the person desiring to know the emotional state.
In one of the implementations of the voice messaging system, the goal was to create an emotion recognizer that can process telephone quality voice messages (8 kHz/8 bit) and can be used as a part of a decision support system for prioritizing voice messages and assigning a person to respond to the message. A classifier was created that can distinguish between two states: “agitation” which includes anger, happiness and fear; and “calm,” which includes normal state and sadness. To create the recognizer, a sampling of 56 telephone messages of varying length (from 15 to 90 seconds) was used. The messages expressed mostly normal and angry emotions that were recorded by eighteen subjects. These utterances were automatically split into 1-3 second segments, which were then evaluated and labeled by persons. The samples were used for creating recognizers using the methodology as described above. A number of ensembles of 15 neural network classifiers for the 8-, 10-, and 14-statistics inputs and the 10- and 20-node architectures were created.
The emotion recognition system is a part of a new generation computerized call center that integrates databases, decision support systems, and different media, such as voice messages, e-mail messages and an Internet server, into one information space. The system consists of three processes: monitoring voice files, distributing voice mail from a voice mail center, and prioritizing messages. Monitoring voice files, which corresponds to the operation 1130 from
The system can also output a statistic of at least one feature or parameter of a voice or voice signal. The statistic may be any of the statistics discussed above, or any other statistic that may be calculated based on the voice signal and its digitization. For example, the length of the entire message may be measured, recorded and displayed, and the percent of silence in the message (an indicator of anger) can be measured, recorded and displayed. The computer system will also contain, in software or firmware, logic for carrying out all of the above tasks as described above. This will include software for measuring, recording and displaying the above features and statistics. This will also include logic and any necessary hardware, such as physical relays and connections, for routing the necessary indications or signals of the detected emotional state, to the desired locations.
Monitoring Telephonic Conversations
The present invention is particularly suited to operation of an emergency response system, such as a 911 system. In such a system, incoming calls are monitored by an embodiment of the present invention. An emotion of the caller would be determined during the caller's conversation with the technician answering the call. The emotion could then be relayed to the emergency response personnel, i.e., police, fire, and/or emergency medical personnel, so they are aware of the emotional state of the caller. In other embodiments, calls may be reviewed and analyzed for better operator performance on future emotional calls.
Operator Performance Evaluation
There are several categories of people for whom emotional training can be beneficial. Among them are autistic people, call center operators and would-be dramatic actors and actresses. Autistic people have problems with understanding emotions and responding adequately to emotional situation. The need for expression of a given emotion in a particular situation should be explained to them, and they should be taught how to react to such a situation. A computer system built as a game can be an ideal patient partner for this purpose. Call center operators need to develop advanced skills in recognizing and portraying emotions in speech. Would-be actors also need to develop such skills. A computerized training program can be used for this purpose.
An indication of the level of emotion or nervousness is 1704 determined and output, preferably before the business event is completed so that one attempting to prevent fraud can make an assessment whether to confront the person before the person leaves. Any kind of display, or output is acceptable, including a paper printout, an audible tone, or a display on a computer screen. This embodiment of the invention may detect emotions other than nervousness. Such emotions include stress or other emotion likely to be displayed by a person committing fraud. The indication of the level of nervousness of the person may be displayed or output in real time to allow one seeking to prevent fraud to obtain results very quickly, so one is able to quickly challenge the person making the suspicious utterance.
As another option, the indication of the level of emotion may include a notification that is sent when the level of emotion or nervousness goes above a predetermined level. The notification may include a visual display on a computer, an auditory sound, etc., or notification to an overseer, the listener, and/or one searching for fraud. The notification could also be sent to a recording device to begin recording the conversation, if the conversation is not already being recorded. The person is then handled 1706 in accordance with the emotion or nervousness detected. In one embodiment, management may develop a set of predetermined responses to help clerks or customer service personnel decide what their course of action should be. The responses may be stored in memory on a CPU 110 of the emotion-detection system, or on a memory accessible to the emotion-detection system, as in ROM 116 of
This embodiment of the present invention has particular application in business areas such as contract negotiations, insurance dealings, customer service, and the like. Fraud in these areas costs companies billions of dollars each year. The invention may also be used in other environments where it may be useful to detect emotions in persons. These may include law enforcement operations, investigations, security checkpoints, building entrances, and the like.
It will be appreciated that a wide range of changes and modifications to the invention as described are contemplated. Accordingly, while preferred embodiments have been shown and described in detail by way of examples, further modifications and embodiments are possible without departing from the scope of the invention as defined by the examples set forth. It is therefore intended that the invention be defined by the claims and all legal equivalents.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US3691652||Jun 1, 1971||Sep 19, 1972||Clynes Manfred E||Programmed system for evoking emotional responses|
|US3855416||Dec 1, 1972||Dec 17, 1974||Fuller F||Method and apparatus for phonation analysis leading to valid truth/lie decisions by fundamental speech-energy weighted vibratto component assessment|
|US3971034||Sep 5, 1972||Jul 20, 1976||Dektor Counterintelligence And Security, Inc.||Physiological response analysis method and apparatus|
|US4093821||Jun 14, 1977||Jun 6, 1978||John Decatur Williamson||Speech analyzer for analyzing pitch or frequency perturbations in individual speech pattern to determine the emotional state of the person|
|US4142067||Apr 11, 1978||Feb 27, 1979||Williamson John D||Speech analyzer for analyzing frequency perturbations in a speech pattern to determine the emotional state of a person|
|US4216594||Oct 30, 1978||Aug 12, 1980||Cheryl Farley||Psychotherapeutic testing game|
|US4472833||Jun 24, 1982||Sep 18, 1984||Turrell Ronald P||Speech aiding by indicating speech rate is excessive|
|US4490840||Mar 30, 1982||Dec 25, 1984||Jones Joseph M||Oral sound analysis method and apparatus for determining voice, speech and perceptual styles|
|US4592086||Dec 8, 1982||May 27, 1986||Nippon Electric Co., Ltd.||Continuous speech recognition system|
|US4602129||Sep 29, 1982||Jul 22, 1986||Vmx, Inc.||Electronic audio communications system with versatile message delivery|
|US4696038||Apr 13, 1983||Sep 22, 1987||Texas Instruments Incorporated||Voice messaging system with unified pitch and voice tracking|
|US4931934||Jun 27, 1988||Jun 5, 1990||Snyder Thomas E||Method and system for measuring clarified intensity of emotion|
|US4996704||Sep 29, 1989||Feb 26, 1991||At&T Bell Laboratories||Electronic messaging systems with additional message storage capability|
|US5163083||Oct 12, 1990||Nov 10, 1992||At&T Bell Laboratories||Automation of telephone operator assistance calls|
|US5410739||Sep 29, 1992||Apr 25, 1995||The Titan Corporation||Variable data message communication over voice communication channel|
|US5461697||Nov 12, 1993||Oct 24, 1995||Sekisui Kagaku Kogyo Kabushiki Kaisha||Speaker recognition system using neural network|
|US5495553||Dec 20, 1993||Feb 27, 1996||Nynex Corporation||Recognizer for recognizing voice messages in pulse code modulated format|
|US5539861||Dec 22, 1993||Jul 23, 1996||At&T Corp.||Speech recognition using bio-signals|
|US5572576||Mar 15, 1994||Nov 5, 1996||Klausner Patent Technologies||Telephone answering device linking displayed data with recorded audio message|
|US5647834||Jun 30, 1995||Jul 15, 1997||Ron; Samuel||Speech-based biofeedback method and system|
|US5666400||Jul 7, 1994||Sep 9, 1997||Bell Atlantic Network Services, Inc.||Intelligent recognition|
|US5704007||Oct 4, 1996||Dec 30, 1997||Apple Computer, Inc.||Utilization of multiple voice sources in a speech synthesizer|
|US5734794||Jun 22, 1995||Mar 31, 1998||White; Tom H.||Method and system for voice-activated cell animation|
|US5774591||Dec 15, 1995||Jun 30, 1998||Xerox Corporation||Apparatus and method for recognizing facial expressions and facial gestures in a sequence of images|
|US5774859||Jan 3, 1995||Jun 30, 1998||Scientific-Atlanta, Inc.||Information system having a speech interface|
|US5812977||Aug 13, 1996||Sep 22, 1998||Applied Voice Recognition L.P.||Voice control computer interface enabling implementation of common subroutines|
|US5860064||Feb 24, 1997||Jan 12, 1999||Apple Computer, Inc.||Method and apparatus for automatic generation of vocal emotion in a synthetic text-to-speech system|
|US5884247||Oct 31, 1996||Mar 16, 1999||Dialect Corporation||Method and apparatus for automated language translation|
|US5893057||Oct 18, 1996||Apr 6, 1999||Ricoh Company Ltd.||Voice-based verification and identification methods and systems|
|US5897616||Jun 11, 1997||Apr 27, 1999||International Business Machines Corporation||Apparatus and methods for speaker verification/identification/classification employing non-acoustic and/or acoustic models and databases|
|US5903870||Sep 18, 1995||May 11, 1999||Vis Tell, Inc.||Voice recognition and display device apparatus and method|
|US5909665||May 29, 1997||Jun 1, 1999||Nec Corporation||Speech recognition system|
|US5913196||Nov 17, 1997||Jun 15, 1999||Talmor; Rita||System and method for establishing identity of a speaker|
|US5918222 *||Mar 15, 1996||Jun 29, 1999||Kabushiki Kaisha Toshiba||Information disclosing apparatus and multi-modal information input/output system|
|US5936515||Apr 15, 1998||Aug 10, 1999||General Signal Corporation||Field programmable voice message device and programming device|
|US5987415||Jun 30, 1998||Nov 16, 1999||Microsoft Corporation||Modeling a user's emotion and personality in a computer user interface|
|US6006188||Mar 19, 1997||Dec 21, 1999||Dendrite, Inc.||Speech signal processing for determining psychological or physiological characteristics using a knowledge base|
|US6151571||Aug 31, 1999||Nov 21, 2000||Andersen Consulting||System, method and article of manufacture for detecting emotion in voice signals through analysis of a plurality of voice signal parameters|
|US6173260||Mar 31, 1998||Jan 9, 2001||Interval Research Corporation||System and method for automatic classification of speech based upon affective content|
|US6212550||Jan 21, 1997||Apr 3, 2001||Motorola, Inc.||Method and system in a client-server for automatically converting messages from a first format to a second format compatible with a message retrieving device|
|US6275806||Aug 31, 1999||Aug 14, 2001||Andersen Consulting, Llp||System method and article of manufacture for detecting emotion in voice signals by utilizing statistics for voice signal parameters|
|US6363346 *||Dec 22, 1999||Mar 26, 2002||Ncr Corporation||Call distribution system inferring mental or physiological state|
|US6542602||Feb 14, 2000||Apr 1, 2003||Nice Systems Ltd.||Telephone call monitoring system|
|US6638217||Dec 16, 1998||Oct 28, 2003||Amir Liberman||Apparatus and methods for detecting emotions|
|US6795808 *||Oct 30, 2000||Sep 21, 2004||Koninklijke Philips Electronics N.V.||User interface/entertainment device that simulates personal interaction and charges external database with relevant data|
|US7181693 *||Mar 17, 2000||Feb 20, 2007||Gateway Inc.||Affective control of information systems|
|US20040002838||Jun 27, 2002||Jan 1, 2004||Oliver Nuria M.||Layered models for context awareness|
|EP1222448B1||Aug 31, 2000||Oct 18, 2006||Accenture LLP||System, method, and article of manufacture for detecting emotion in voice signals by utilizing statistics for voice signal parameters|
|EP1222656B1||Aug 31, 2000||Jun 15, 2005||Accenture LLP||Telephonic emotion detector with operator feedback|
|WO1987002491A1||Oct 9, 1986||Apr 23, 1987||Victor Campbell Blackwell||Personal identification device|
|WO1998003941A1||Jul 24, 1997||Jan 29, 1998||Chiptec International Ltd.||Identity card, information carrier and housing designed for its application|
|WO1998010412A2||Sep 9, 1997||Mar 12, 1998||Voice Control Systems, Inc.||Speech verification system and secure data transmission|
|WO1998015924A2||Sep 29, 1997||Apr 16, 1998||Smarttouch||Tokenless biometric automated teller machine access system|
|WO1998023062A1||Nov 21, 1997||May 28, 1998||T-Netix, Inc.||Voice recognition for information system access and transaction processing|
|WO1999022364A1||Oct 29, 1998||May 6, 1999||Interval Research Corporation||System and method for automatically classifying the affective content of speech|
|WO1999031653A1||Dec 16, 1998||Jun 24, 1999||Carmel, Avi||Apparatus and methods for detecting emotions|
|WO2000062279A1||Apr 11, 2000||Oct 19, 2000||Amir Liberman||Apparatus and methods for detecting emotions in the human voice|
|WO2001016570A1||Aug 31, 2000||Mar 8, 2001||Accenture Llp||System, method, and article of manufacture for detecting emotion in voice signals by utilizing statistics for voice signal parameters|
|WO2001016936A1||Aug 31, 2000||Mar 8, 2001||Accenture Llp||Voice recognition for internet navigation|
|WO2001016938A1||Aug 31, 2000||Mar 8, 2001||Accenture Llp||A system, method, and article of manufacture for detecting emotion in voice signals through analysis of a plurality of voice signal parameters|
|WO2001016940A1||Aug 31, 2000||Mar 8, 2001||Accenture, Llp||System, method, and article of manufacture for a voice recognition system for identity authentication in order to gain access to data on the internet|
|1||Banse, Rainer, et al., "Acoustic Profiles in Vocal Emotion Express," Journal of Personality and Social Psychology, 1996, vol. 70, No. 3, pp. 614-636.|
|2||Boersma, Paul, "Accurate Short-Term Analysis of the Fundamental Frequency and the Harmonics-to-Noise Ratio of a Sampled Sound," Institute of Phonetic Sciences, University of Amsterdam, Proceedings 17 (1993), pp. 97-110.|
|3||Bourjot, C., et al., "Phonetic Decoder Assessment," Eurospeech 89, European Conference on Speech Communication and Technology, vol. Two, Sep. 1989, pp. 457-460.|
|4||Breiman, Leo, "Bagging Predictors," Machine Learning, 24, pp. 23-140 (1996).|
|5||Cahn, Janet E., The Generation of Affect in Synthesized Speech, M.I.T. Media Technology Laboratory, 1990, pp. 1-19.|
|6||Campbell, Jr., Joseph P., et al., Government Applications and Operations, Biometric Consortium, http://www.biometrics.org/REPORTS/CTSTG96, 1996.|
|7||Chiu, C.C., et al., "The Analysis and Recognition of Human Vocal Emotions," Proceedings of International Computer Symposium, Dec. 1994, pp. 83-88.|
|8||Cowie, R., et al., "Emotion Recognition in Human-Computer Interaction," IEEE Signal Processing Magazine, Jan. 2001, pp. 33-80.|
|9||Darby, M.D., John K., editor, Speech Evaluation in Psychiatry, Chapter 109, 1981, pp. 189-220.|
|10||Dellaert, Frank, et al., "Recognizing Emotion in Speech," 1996, ICSLP 96, Proceedings, Fourth International Conference on Spoken Language, vol. 3, pp. 1970-1973.|
|11||Elliott, Clark, et al., Autonomous Agents as Synthetic Characters, American Association for Artificial Intelligence, Summer 1998, pp. 13-30.|
|12||Foreign communication/office action dated Dec. 3, 2008 (in Hebrew and English translation citing above U.S. Patent No. 5,572,576) (5p).|
|13||Gadallah, Mahmoud E., et al., "Speech Based Automatic Lie Detection," 16th National Radio Science Conference, NRSC'99, Ain Shams University, Feb. 23-25, 1999, pp. C33/1-C33/8.|
|14||Hansen, John H.L., Analysis and compensation of speech under stress and noise for environmental robustness in speech recognition, Speech Communication 20 (1996), pp. 151-173.|
|15||Hansen, Lars Kai, "Neutral Network Ensembles," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 12, No. 10, Oct. 1990, pp. 993-1001.|
|16||Hays, Ronald J., INS Passenger Accelerated Service System (INSPASS), Biometric Consortium; http://www.biometrics.org/REPORTS/INSASS.html, 1996.|
|17||Jimenez-Fernandez, Alfonso, et al., "Pattern Recognition in the Vocal Expression of Emotional Categories," IEEE Ninth Annual Conference of the Engineering in Medicine and Biology Society, 1987, pp. 2090-2091.|
|18||Klasmeyer, Gudrun, "The Perceptual Importance of Selected Voice Quality Parameters," IEEE International Conference on Acoustics, Speech and Signal Processing, vol. 3, Apr. 21-24, 1997, pp. 1615-1618.|
|19||Kononenko, Igor, Estimating Attributes: Analysis and Extensions of RELIEF, University of Ljubljana, Faculty of Electrical Engineering & Computer Science, 1994, pp. 171-182.|
|20||Lee, Chul Min, et al., "Toward Detecting Emotions in Spoken Dialogs", IEEE Transactions on Speech and Audio Processing, vol. 13, No. 2, Mar. 2005, pp. 293-303.|
|21||Makarova, V. and Petrushin, V.A., "Vowel Quality in Emotional Speech," Proc. 10th International Conference on Speech and Computer, Patras, Greece, pp. 449-454 (2005).|
|22||Mayer, David L., et al., "Development of a Speech Analysis Protocol for Accident Investigation," Abstract and Proceedings of the 38th Annual Meeting of the Human Factors and Ergonomic Society, vol. 1, Oct. 24-28, 1994, pp. 124-127.|
|23||Moriyama, Tsuyoshi, et al., "Emotion Recognition and Synthesis System on Speech," 1999 IEEE, pp. 840-844.|
|24||Murray, Iain R., et al., "Toward the simulation of emotion in synthetic speech: A review of the literature on human vocal emotion," J. Acoust. Soc. Am. 93 (2), Feb. 1993, pp. 1097-1108.|
|25||Oliver, Gina M., A Study of the Use of Biometrics as it Relates to Personal Privacy Concerns, 1999, http://nile.ed.umuc.edu/~jmeinke/inss690/oliver/Oliver-690.html.|
|26||Parmanto, Bambang, et al., Improving Committee Diagnosis with Resampling Techniques, Department of Information Science, University of Pittsburgh, 1996, pp. 882-888.|
|27||*||Petrushin, V. "Emotion in Speech: Recognition and Application to Call Centers," Artificial Neural Networks in Engineering, Nov. 7-10, 1999.|
|28||Petrushin, V.A. and Makarova, V., Parameters of Fricatives and Affricates in Russian Emotional Speech, Proc. 11th International Conference on Speech and Computer, Saint-Petersburg, Russia, pp. 73-80 (2006).|
|29||Polzin, Thomas S., et al., "Detecting Emotions in Speech," Proceedings of the CMC 1998, School of Computer Science, Carnegie Mellon University, Pittsburgh, http://www.ri.cmu.edu/pubs/pub-2161.html.|
|30||Polzin, Thomas S., et al., "Pronunciation Variations in Emotional Speech," ESCA-98, Tutorial and Research Workshop on Modeling Pronunciation Variation for Automatic Speech Recognition, May 1998, Interactive Systems Laboratories, Carnegie Mellon University, Pittsburgh, http://www.ri.cmu.edu/pubs/pub-2160.html.|
|31||*||Savitch, Walter. "Problem Solving with C++: The Object of Programming" Addison Wesley Longman, Inc. pp. 2-7, 1999.|
|32||Scherer, Klaus R., et al., "Vocal Cues in Emotion Encoding and Decoding," Motivation and Emotion, vol. 15, No. 2, 1991, pp. 123-148.|
|33||Slaney, Malcolm, et al., "Baby Ears: a recognition system for affective vocalizations," 1998 IEEE, International Conference on Acoustics, Speech and Signal Processing, vol. 2, pp. 985-988.|
|34||Talbot, David, "Prosody," Technology Review, Jul./Aug. 2002, p. 27.|
|35||Tosa, Naoko, et al., Life-like Communication Agent, MIC & MUSE, 1996, pp. 1-15, http://www.mic.atr.co.jp/~tosa.|
|36||Webster's New World Dictionary of American English (Third College Ed. 1988) definition of "pre" (relied on by the Examiner in U.S. Appl. No. 09/833,301), p. 1060.|
|37||Yamada, Toyotoshi, et al., "Pattern recognition of emotion with Neural Network," 1995 IEEE, Proceedings of IECON 21st International Conference on Industrial Electronics, Control and Instrumentation, vol. 1, pp. 183-187.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7949103 *||Jun 20, 2007||May 24, 2011||Vontoo, Llc||System and method for providing voice messaging with dynamic content|
|US8078465||Dec 13, 2011||Lena Foundation||System and method for detection and analysis of speech|
|US8204747 *||May 21, 2007||Jun 19, 2012||Panasonic Corporation||Emotion recognition apparatus|
|US8392195 *||Mar 5, 2013||International Business Machines Corporation||Multiple audio/video data stream simulation|
|US8493410||May 1, 2012||Jul 23, 2013||International Business Machines Corporation||Simulation method and system|
|US8559469 *||Feb 8, 2012||Oct 15, 2013||Open Invention Network, Llc||System and method for voice transmission over network protocols|
|US8644550||May 31, 2012||Feb 4, 2014||International Business Machines Corporation||Multiple audio/video data stream simulation|
|US8675846||Apr 15, 2011||Mar 18, 2014||Rpx Corporation||System and method for providing voice messaging with dynamic content|
|US8744847||Apr 25, 2008||Jun 3, 2014||Lena Foundation||System and method for expressive language assessment|
|US8781880||Jun 4, 2013||Jul 15, 2014||Rank Miner, Inc.||System, method and apparatus for voice analytics of recorded audio|
|US8825479 *||Oct 24, 2013||Sep 2, 2014||Simple Emotion, Inc.||System and method for recognizing emotional state from a speech signal|
|US8886528 *||Jun 2, 2010||Nov 11, 2014||Panasonic Corporation||Audio signal processing device and method|
|US8938390||Feb 27, 2009||Jan 20, 2015||Lena Foundation||System and method for expressive language and developmental disorder assessment|
|US9191513||Aug 18, 2014||Nov 17, 2015||Wipro Limited||System and method for dynamic job allocation based on acoustic sentiments|
|US9224402 *||Sep 30, 2013||Dec 29, 2015||International Business Machines Corporation||Wideband speech parameterization for high quality synthesis, transformation and quantization|
|US9240188||Jan 23, 2009||Jan 19, 2016||Lena Foundation||System and method for expressive language, developmental disorder, and emotion assessment|
|US9280914||Apr 10, 2014||Mar 8, 2016||National Central University||Vision-aided hearing assisting device|
|US9294814||May 24, 2013||Mar 22, 2016||International Business Machines Corporation||Simulation method and system|
|US9311546 *||Nov 27, 2009||Apr 12, 2016||Nottingham Trent University||Biometric identity verification for access control using a trained statistical classifier|
|US9355651||Apr 29, 2014||May 31, 2016||Lena Foundation||System and method for expressive language, developmental disorder, and emotion assessment|
|US20080040110 *||Aug 8, 2005||Feb 14, 2008||Nice Systems Ltd.||Apparatus and Methods for the Detection of Emotions in Audio Interactions|
|US20080123822 *||Jun 20, 2007||May 29, 2008||Dustin Kenneth Sapp||System and method for providing voice messaging with dynamic content|
|US20080235016 *||Jan 23, 2008||Sep 25, 2008||Infoture, Inc.||System and method for detection and analysis of speech|
|US20080249764 *||Dec 5, 2007||Oct 9, 2008||Microsoft Corporation||Smart Sentiment Classifier for Product Reviews|
|US20090175424 *||Jan 5, 2009||Jul 9, 2009||Siemens Aktiengesellschaft Of Munich, Germany||Method for providing service for user|
|US20090313019 *||May 21, 2007||Dec 17, 2009||Yumiko Kato||Emotion recognition apparatus|
|US20110222672 *||Sep 15, 2011||Dustin Kenneth Sapp||System and method for providing voice messaging with dynamic content|
|US20110285504 *||Nov 27, 2009||Nov 24, 2011||Sergio Grau Puerto||Biometric identity verification|
|US20120089393 *||Jun 2, 2010||Apr 12, 2012||Naoya Tanaka||Acoustic signal processing device and method|
|US20120239393 *||Sep 20, 2012||International Business Machines Corporation||Multiple audio/video data stream simulation|
|US20140052448 *||Oct 24, 2013||Feb 20, 2014||Simple Emotion, Inc.||System and method for recognizing emotional state from a speech signal|
|US20150081299 *||Jun 1, 2012||Mar 19, 2015||Koninklijke Philips N.V.||Method and system for assisting patients|
|US20150095035 *||Sep 30, 2013||Apr 2, 2015||International Business Machines Corporation||Wideband speech parameterization for high quality synthesis, transformation and quantization|
|US20150170674 *||Dec 9, 2014||Jun 18, 2015||Sony Corporation||Information processing apparatus, information processing method, and program|
|WO2010085681A1 *||Jan 22, 2010||Jul 29, 2010||Infoture, Inc.||System and method for expressive language, developmental disorder, and emotion assessment|
|U.S. Classification||704/270, 704/274, 704/231|
|International Classification||G10L17/00, G10L21/00, G10L15/00|
|Cooperative Classification||G10L25/30, G10L17/26|
|Jun 18, 2010||AS||Assignment|
Owner name: ACCENTURE LLP,CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:PETRUSHIN, VALERY A.;REEL/FRAME:024555/0652
Effective date: 20020712
|Sep 8, 2010||AS||Assignment|
Owner name: ACCENTURE GLOBAL SERVICES GMBH, SWITZERLAND
Free format text: CONFIRMATORY ASSIGNMENT;ASSIGNOR:ACCENTURE LLP;REEL/FRAME:024946/0971
Effective date: 20100831
|Jan 26, 2011||AS||Assignment|
Owner name: ACCENTURE GLOBAL SERVICES LIMITED, IRELAND
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:ACCENTURE GLOBAL SERVICES GMBH;REEL/FRAME:025700/0287
Effective date: 20100901
|Mar 4, 2013||FPAY||Fee payment|
Year of fee payment: 4